ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.26 by root, Fri Nov 22 09:40:13 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_FORBID_OBJECTS 0x00000010UL
105#define F_PACK_STRINGS 0x00000020UL
106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
108#define F_VALIDATE_UTF8 0x00000100UL
63 109
64#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
65 111
66#define SB do { 112#define SB do {
67#define SE } while (0) 113#define SE } while (0)
79# define CBOR_SLOW 0 125# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 126# define CBOR_STASH cbor_stash
81#endif 127#endif
82 128
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 129static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 130static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 131
86typedef struct { 132typedef struct {
87 U32 flags; 133 U32 flags;
88 U32 max_depth; 134 U32 max_depth;
89 STRLEN max_size; 135 STRLEN max_size;
136 SV *filter;
137
138 // for the incremental parser
139 STRLEN incr_pos; // the current offset into the text
140 STRLEN incr_need; // minimum bytes needed to decode
141 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
90} CBOR; 142} CBOR;
91 143
92ecb_inline void 144ecb_inline void
93cbor_init (CBOR *cbor) 145cbor_init (CBOR *cbor)
94{ 146{
95 Zero (cbor, 1, CBOR); 147 Zero (cbor, 1, CBOR);
96 cbor->max_depth = 512; 148 cbor->max_depth = 512;
149}
150
151ecb_inline void
152cbor_free (CBOR *cbor)
153{
154 SvREFCNT_dec (cbor->filter);
155 SvREFCNT_dec (cbor->incr_count);
97} 156}
98 157
99///////////////////////////////////////////////////////////////////////////// 158/////////////////////////////////////////////////////////////////////////////
100// utility functions 159// utility functions
101 160
131{ 190{
132 return idx > 23 191 return idx > 23
133 ? idx > 0xffU 192 ? idx > 0xffU
134 ? idx > 0xffffU 193 ? idx > 0xffffU
135 ? idx > 0xffffffffU 194 ? idx > 0xffffffffU
195 ? 11
136 ? 7 196 : 7
137 : 6
138 : 5 197 : 5
139 : 4 198 : 4
140 : 3; 199 : 3;
141} 200}
142 201
179static void 238static void
180encode_uint (enc_t *enc, int major, UV len) 239encode_uint (enc_t *enc, int major, UV len)
181{ 240{
182 need (enc, 9); 241 need (enc, 9);
183 242
184 if (len < 24) 243 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
186 else if (len <= 0xff) 245 else if (ecb_expect_true (len <= 0xffU))
187 { 246 {
188 *enc->cur++ = major | 24; 247 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 248 *enc->cur++ = len;
190 } 249 }
191 else if (len <= 0xffff) 250 else if (len <= 0xffffU)
192 { 251 {
193 *enc->cur++ = major | 25; 252 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 254 *enc->cur++ = len;
196 } 255 }
197 else if (len <= 0xffffffff) 256 else if (len <= 0xffffffffU)
198 { 257 {
199 *enc->cur++ = major | 26; 258 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 262 *enc->cur++ = len;
204 } 263 }
205 else 264 else
206 { 265 {
207 *enc->cur++ = major | 27; 266 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 267 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 268 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 269 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 270 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 271 *enc->cur++ = len >> 24;
217} 276}
218 277
219ecb_inline void 278ecb_inline void
220encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
221{ 280{
222 encode_uint (enc, 0xc0, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
223} 282}
224 283
225static void 284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 287{
288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
307ecb_inline void
308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
318 need (enc, len);
319 memcpy (enc->cur, str, len);
320 enc->cur += len;
321}
322
323ecb_inline void
324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
325{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 327 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 329
232 if (SvOK (*svp)) 330 if (SvOK (*svp))
233 { 331 {
234 // already registered, use stringref 332 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 333 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 334 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 335 return;
238 } 336 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 337 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 338 {
241 // register only 339 // register only
242 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 341 ++enc->stringref_idx;
244 } 342 }
245 } 343 }
246 344
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
248 need (enc, len);
249 memcpy (enc->cur, str, len);
250 enc->cur += len;
251} 346}
252 347
253static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
254 349
255static void 350static void
260 if (enc->depth >= enc->cbor.max_depth) 355 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 356 croak (ERR_NESTING_EXCEEDED);
262 357
263 ++enc->depth; 358 ++enc->depth;
264 359
265 encode_uint (enc, 0x80, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 361
362 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
268 { 364 {
269 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
272 374
273 --enc->depth; 375 --enc->depth;
274} 376}
275 377
276static void 378static void
285 387
286 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
288 390
289 if (mg) 391 if (mg)
290 encode_ch (enc, 0xa0 | 31); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 393 else
292 encode_uint (enc, 0xa0, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
293 395
294 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
295 { 397 {
296 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
298 else 400 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 402
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 404 }
303 405
304 if (mg) 406 if (mg)
305 encode_ch (enc, 0xe0 | 31); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 408
307 --enc->depth; 409 --enc->depth;
308} 410}
309 411
310// encode objects, arrays and special \0=false and \1=true values. 412// encode objects, arrays and special \0=false and \1=true values.
311static void 413static void
312encode_rv (enc_t *enc, SV *sv) 414encode_rv (enc_t *enc, SV *sv)
313{ 415{
314 SvGETMAGIC (sv); 416 SvGETMAGIC (sv);
315
316 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317 && ecb_expect_false (SvREFCNT (sv) > 1))
318 {
319 if (!enc->shareable)
320 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324 if (SvOK (*svp))
325 {
326 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327 encode_uint (enc, 0x00, SvUV (*svp));
328 return;
329 }
330 else
331 {
332 sv_setuv (*svp, enc->shareable_idx);
333 ++enc->shareable_idx;
334 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335 }
336 }
337 417
338 svtype svt = SvTYPE (sv); 418 svtype svt = SvTYPE (sv);
339 419
340 if (ecb_expect_false (SvOBJECT (sv))) 420 if (ecb_expect_false (SvOBJECT (sv)))
341 { 421 {
348 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 428 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349 ? cbor_tagged_stash 429 ? cbor_tagged_stash
350 : gv_stashpv ("CBOR::XS::Tagged" , 1); 430 : gv_stashpv ("CBOR::XS::Tagged" , 1);
351 431
352 HV *stash = SvSTASH (sv); 432 HV *stash = SvSTASH (sv);
353 GV *method;
354 433
355 if (stash == boolean_stash) 434 if (stash == boolean_stash)
356 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 435 {
436 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
437 return;
438 }
357 else if (stash == error_stash) 439 else if (stash == error_stash)
358 encode_ch (enc, 0xe0 | 23); 440 {
441 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
442 return;
443 }
359 else if (stash == tagged_stash) 444 else if (stash == tagged_stash)
360 { 445 {
361 if (svt != SVt_PVAV) 446 if (svt != SVt_PVAV)
362 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 447 croak ("encountered CBOR::XS::Tagged object that isn't an array");
363 448
364 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 449 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 450 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
451
452 return;
453 }
454 }
455
456 if (ecb_expect_false (SvREFCNT (sv) > 1)
457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
458 {
459 if (!enc->shareable)
460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
461
462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
463
464 if (SvOK (*svp))
366 } 465 {
466 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
467 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
468 return;
469 }
470 else
471 {
472 sv_setuv (*svp, enc->shareable_idx);
473 ++enc->shareable_idx;
474 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
475 }
476 }
477
478 if (ecb_expect_false (SvOBJECT (sv)))
479 {
480 HV *stash = SvSTASH (sv);
481 GV *method;
482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
367 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368 { 487 {
369 dSP; 488 dSP;
370 489
371 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
372 // we re-bless the reference to get overload and other niceties right 492 // we re-bless the reference to get overload and other niceties right
373 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
374 494
375 PUTBACK; 495 PUTBACK;
376 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
389 } 509 }
390 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
391 { 511 {
392 dSP; 512 dSP;
393 513
394 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
395 EXTEND (SP, 2); 517 EXTEND (SP, 2);
396 // we re-bless the reference to get overload and other niceties right 518 // we re-bless the reference to get overload and other niceties right
397 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
398 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
399 521
404 // catch this surprisingly common error 526 // catch this surprisingly common error
405 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
407 529
408 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409 encode_uint (enc, 0x80, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
410 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411 533
412 while (count) 534 while (count)
413 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
414 536
415 PUTBACK; 537 PUTBACK;
436{ 558{
437 double nv = SvNVX (sv); 559 double nv = SvNVX (sv);
438 560
439 need (enc, 9); 561 need (enc, 9);
440 562
441 if (ecb_expect_false (nv == (U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
442 encode_uint (enc, 0x00, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443 //TODO: maybe I32? 565 //TODO: maybe I32?
444 else if (ecb_expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
445 { 567 {
446 uint32_t fp = ecb_float_to_binary32 (nv); 568 uint32_t fp = ecb_float_to_binary32 (nv);
447 569
448 *enc->cur++ = 0xe0 | 26; 570 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
449 571
450 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
451 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
452 574
453 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
455 } 577 }
456 else 578 else
457 { 579 {
458 uint64_t fp = ecb_double_to_binary64 (nv); 580 uint64_t fp = ecb_double_to_binary64 (nv);
459 581
460 *enc->cur++ = 0xe0 | 27; 582 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
461 583
462 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
463 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
464 586
465 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
474 596
475 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
476 { 598 {
477 STRLEN len; 599 STRLEN len;
478 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
479 encode_str (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
480 } 602 }
481 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
482 encode_nv (enc, sv); 604 encode_nv (enc, sv);
483 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
484 { 606 {
485 if (SvIsUV (sv)) 607 if (SvIsUV (sv))
486 encode_uint (enc, 0x00, SvUVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
487 else if (SvIVX (sv) >= 0) 609 else if (SvIVX (sv) >= 0)
488 encode_uint (enc, 0x00, SvIVX (sv)); 610 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
489 else 611 else
490 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 612 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
491 } 613 }
492 else if (SvROK (sv)) 614 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 615 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 616 else if (!SvOK (sv))
495 encode_ch (enc, 0xe0 | 22); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 618 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497 encode_ch (enc, 0xe0 | 23); 619 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498 else 620 else
499 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 621 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
500 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 622 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501} 623}
502 624
503static SV * 625static SV *
504encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
505{ 627{
506 enc_t enc = { }; 628 enc_t enc = { 0 };
507 629
508 enc.cbor = *cbor; 630 enc.cbor = *cbor;
509 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
510 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
511 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
512 634
513 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
514 636
515 if (cbor->flags & F_ALLOW_STRINGREF) 637 if (cbor->flags & F_PACK_STRINGS)
516 { 638 {
517 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 639 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 640 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 641 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520 } 642 }
542 CBOR cbor; 664 CBOR cbor;
543 U32 depth; // recursion depth 665 U32 depth; // recursion depth
544 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
545 AV *shareable; 667 AV *shareable;
546 AV *stringref; 668 AV *stringref;
669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
547} dec_t; 671} dec_t;
548 672
549#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
550 680
551#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
552 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703ecb_cold static void
704err_nesting_exceeded (dec_t *dec)
705{
706 err_set (dec, ERR_NESTING_EXCEEDED);
707}
708
709#define ERR_DO(do) SB do; goto fail; SE
710#define ERR(reason) ERR_DO (err_set (dec, reason))
711#define ERR_ERRSV ERR_DO (err_errsv (dec))
712
713#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
714
553#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 715#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
554#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
555 717
556static UV 718static UV
557decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
558{ 720{
559 switch (*dec->cur & 31) 721 U8 m = *dec->cur & MINOR_MASK;
560 { 722 ++dec->cur;
561 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564 return *dec->cur++ & 31;
565 723
566 case 24: 724 if (ecb_expect_true (m < LENGTH_EXT1))
725 return m;
726 else if (ecb_expect_true (m == LENGTH_EXT1))
727 {
567 WANT (2); 728 WANT (1);
568 dec->cur += 2; 729 dec->cur += 1;
569 return dec->cur[-1]; 730 return dec->cur[-1];
570 731 }
571 case 25: 732 else if (ecb_expect_true (m == LENGTH_EXT2))
733 {
572 WANT (3); 734 WANT (2);
573 dec->cur += 3; 735 dec->cur += 2;
574 return (((UV)dec->cur[-2]) << 8) 736 return (((UV)dec->cur[-2]) << 8)
575 | ((UV)dec->cur[-1]); 737 | ((UV)dec->cur[-1]);
576 738 }
577 case 26: 739 else if (ecb_expect_true (m == LENGTH_EXT4))
740 {
578 WANT (5); 741 WANT (4);
579 dec->cur += 5; 742 dec->cur += 4;
580 return (((UV)dec->cur[-4]) << 24) 743 return (((UV)dec->cur[-4]) << 24)
581 | (((UV)dec->cur[-3]) << 16) 744 | (((UV)dec->cur[-3]) << 16)
582 | (((UV)dec->cur[-2]) << 8) 745 | (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 746 | ((UV)dec->cur[-1]);
584 747 }
585 case 27: 748 else if (ecb_expect_true (m == LENGTH_EXT8))
749 {
586 WANT (9); 750 WANT (8);
587 dec->cur += 9; 751 dec->cur += 8;
752
753 return
754#if UVSIZE < 8
755 0
756#else
588 return (((UV)dec->cur[-8]) << 56) 757 (((UV)dec->cur[-8]) << 56)
589 | (((UV)dec->cur[-7]) << 48) 758 | (((UV)dec->cur[-7]) << 48)
590 | (((UV)dec->cur[-6]) << 40) 759 | (((UV)dec->cur[-6]) << 40)
591 | (((UV)dec->cur[-5]) << 32) 760 | (((UV)dec->cur[-5]) << 32)
761#endif
592 | (((UV)dec->cur[-4]) << 24) 762 | (((UV)dec->cur[-4]) << 24)
593 | (((UV)dec->cur[-3]) << 16) 763 | (((UV)dec->cur[-3]) << 16)
594 | (((UV)dec->cur[-2]) << 8) 764 | (((UV)dec->cur[-2]) << 8)
595 | ((UV)dec->cur[-1]); 765 | ((UV)dec->cur[-1]);
596 766 }
597 default: 767 else
598 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 768 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599 }
600 769
601fail: 770fail:
602 return 0; 771 return 0;
603} 772}
604 773
609{ 778{
610 AV *av = newAV (); 779 AV *av = newAV ();
611 780
612 DEC_INC_DEPTH; 781 DEC_INC_DEPTH;
613 782
614 if ((*dec->cur & 31) == 31) 783 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615 { 784 {
616 ++dec->cur; 785 ++dec->cur;
617 786
618 for (;;) 787 for (;;)
619 { 788 {
620 WANT (1); 789 WANT (1);
621 790
622 if (*dec->cur == (0xe0 | 31)) 791 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
623 { 792 {
624 ++dec->cur; 793 ++dec->cur;
625 break; 794 break;
626 } 795 }
627 796
628 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
629 } 798 }
630 } 799 }
631 else 800 else
632 { 801 {
633 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
634 803
804 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635 av_fill (av, len - 1); 805 av_fill (av, len - 1);
636 806
637 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
638 AvARRAY (av)[i] = decode_sv (dec); 808 AvARRAY (av)[i] = decode_sv (dec);
639 } 809 }
652{ 822{
653 // for speed reasons, we specialcase single-string 823 // for speed reasons, we specialcase single-string
654 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
655 825
656 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
657 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658 { 828 {
659 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
660 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
661 831
832 WANT (len);
662 dec->cur += len; 833 dec->cur += len;
663 834
664 if (ecb_expect_false (dec->stringref))
665 av_push (dec->stringref, newSVpvn (key, len));
666
667 hv_store (hv, key, len, decode_sv (dec), 0); 835 hv_store (hv, key, len, decode_sv (dec), 0);
668 836
669 return; 837 return;
670 } 838 }
671 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 839 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672 { 840 {
673 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
674 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
675 843
844 WANT (len);
676 dec->cur += len; 845 dec->cur += len;
677 846
678 if (ecb_expect_false (dec->stringref)) 847 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 848 if (!is_utf8_string (key, len))
849 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
680 850
681 hv_store (hv, key, -len, decode_sv (dec), 0); 851 hv_store (hv, key, -len, decode_sv (dec), 0);
682 852
683 return; 853 return;
684 } 854 }
685 855
686 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
687 SV *v = decode_sv (dec); 857 SV *v = decode_sv (dec);
688 858
859 // we leak memory if uncaught exceptions are thrown by random magical
860 // methods, and this is hopefully the only place where it can happen,
861 // so if there is a chance of an exception, take the very slow path.
862 // since catching exceptions is "undocumented/internal/forbidden" by
863 // the new p5p powers, we need to call out to a perl function :/
864 if (ecb_expect_false (SvAMAGIC (k)))
865 {
866 dSP;
867
868 ENTER; SAVETMPS;
869 PUSHMARK (SP);
870 EXTEND (SP, 3);
871 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
872 PUSHs (sv_2mortal (k));
873 PUSHs (sv_2mortal (v));
874
875 PUTBACK;
876 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
877 SPAGAIN;
878
879 FREETMPS; LEAVE;
880
881 if (SvTRUE (ERRSV))
882 ERR_ERRSV;
883
884 return;
885 }
886
689 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
690 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
889
890fail:
891 ;
691} 892}
692 893
693static SV * 894static SV *
694decode_hv (dec_t *dec) 895decode_hv (dec_t *dec)
695{ 896{
696 HV *hv = newHV (); 897 HV *hv = newHV ();
697 898
698 DEC_INC_DEPTH; 899 DEC_INC_DEPTH;
699 900
700 if ((*dec->cur & 31) == 31) 901 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701 { 902 {
702 ++dec->cur; 903 ++dec->cur;
703 904
704 for (;;) 905 for (;;)
705 { 906 {
706 WANT (1); 907 WANT (1);
707 908
708 if (*dec->cur == (0xe0 | 31)) 909 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
709 { 910 {
710 ++dec->cur; 911 ++dec->cur;
711 break; 912 break;
712 } 913 }
713 914
714 decode_he (dec, hv); 915 decode_he (dec, hv);
715 } 916 }
716 } 917 }
717 else 918 else
718 { 919 {
719 int pairs = decode_uint (dec); 920 UV pairs = decode_uint (dec);
921
922 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
720 923
721 while (pairs--) 924 while (pairs--)
722 decode_he (dec, hv); 925 decode_he (dec, hv);
723 } 926 }
724 927
734static SV * 937static SV *
735decode_str (dec_t *dec, int utf8) 938decode_str (dec_t *dec, int utf8)
736{ 939{
737 SV *sv = 0; 940 SV *sv = 0;
738 941
739 if ((*dec->cur & 31) == 31) 942 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
740 { 943 {
944 // indefinite length strings
741 ++dec->cur; 945 ++dec->cur;
742 946
947 U8 major = *dec->cur & MAJOR_MISC;
948
743 sv = newSVpvn ("", 0); 949 sv = newSVpvn ("", 0);
744 950
745 // not very fast, and certainly not robust against illegal input
746 for (;;) 951 for (;;)
747 { 952 {
748 WANT (1); 953 WANT (1);
749 954
750 if (*dec->cur == (0xe0 | 31)) 955 if ((*dec->cur - major) > LENGTH_EXT8)
956 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751 { 957 {
752 ++dec->cur; 958 ++dec->cur;
753 break; 959 break;
754 } 960 }
961 else
962 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755 963
756 sv_catsv (sv, decode_sv (dec)); 964 STRLEN len = decode_uint (dec);
965
966 WANT (len);
967 sv_catpvn (sv, dec->cur, len);
968 dec->cur += len;
757 } 969 }
758 } 970 }
759 else 971 else
760 { 972 {
761 STRLEN len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
768 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 980 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 981 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 982 }
771 983
772 if (utf8) 984 if (utf8)
985 {
986 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
987 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
988 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
989
773 SvUTF8_on (sv); 990 SvUTF8_on (sv);
991 }
774 992
775 return sv; 993 return sv;
776 994
777fail: 995fail:
778 SvREFCNT_dec (sv); 996 SvREFCNT_dec (sv);
797 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
798 break; 1016 break;
799 1017
800 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
801 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
802 ENTER; SAVETMPS; 1022 ENTER;
803 1023
804 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
805 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
806 1026
807 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
808 1028
809 FREETMPS; LEAVE; 1029 LEAVE;
810 } 1030 }
811 break; 1031 break;
812 1032
813 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
814 { 1034 {
815 if ((*dec->cur >> 5) != 0) 1035 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1036 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817 1037
818 UV idx = decode_uint (dec); 1038 UV idx = decode_uint (dec);
819 1039
820 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1040 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
827 case CBOR_TAG_VALUE_SHAREABLE: 1047 case CBOR_TAG_VALUE_SHAREABLE:
828 { 1048 {
829 if (ecb_expect_false (!dec->shareable)) 1049 if (ecb_expect_false (!dec->shareable))
830 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1050 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831 1051
1052 if (dec->cbor.flags & F_ALLOW_CYCLES)
1053 {
832 sv = newSV (0); 1054 sv = newSV (0);
833 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1055 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
834 1056
835 SV *osv = decode_sv (dec); 1057 SV *osv = decode_sv (dec);
836 sv_setsv (sv, osv); 1058 sv_setsv (sv, osv);
837 SvREFCNT_dec_NN (osv); 1059 SvREFCNT_dec_NN (osv);
1060 }
1061 else
1062 {
1063 av_push (dec->shareable, &PL_sv_undef);
1064 int idx = AvFILLp (dec->shareable);
1065 sv = decode_sv (dec);
1066 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1067 }
838 } 1068 }
839 break; 1069 break;
840 1070
841 case CBOR_TAG_VALUE_SHAREDREF: 1071 case CBOR_TAG_VALUE_SHAREDREF:
842 { 1072 {
843 if ((*dec->cur >> 5) != 0) 1073 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1074 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845 1075
846 UV idx = decode_uint (dec); 1076 UV idx = decode_uint (dec);
847 1077
848 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1078 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
849 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1079 ERR ("corrupted CBOR data (sharedref index out of bounds)");
850 1080
851 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1081 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1082
1083 if (sv == &PL_sv_undef)
1084 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
852 } 1085 }
853 break; 1086 break;
854 1087
855 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
856 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
857 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
858 1094
859 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
860 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
861 1097
871 if (!method) 1107 if (!method)
872 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1108 ERR ("cannot decode perl-object (package does not have a THAW method)");
873 1109
874 dSP; 1110 dSP;
875 1111
876 ENTER; SAVETMPS; PUSHMARK (SP); 1112 ENTER; SAVETMPS;
1113 PUSHMARK (SP);
877 EXTEND (SP, len + 1); 1114 EXTEND (SP, len + 1);
878 // we re-bless the reference to get overload and other niceties right 1115 // we re-bless the reference to get overload and other niceties right
879 PUSHs (*av_fetch (av, 0, 1)); 1116 PUSHs (*av_fetch (av, 0, 1));
880 PUSHs (sv_cbor); 1117 PUSHs (sv_cbor);
881 1118
889 SPAGAIN; 1126 SPAGAIN;
890 1127
891 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
892 { 1129 {
893 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
894 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
895 } 1132 }
896 1133
897 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
898 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
899 1136
902 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
903 } 1140 }
904 break; 1141 break;
905 1142
906 default: 1143 default:
1144 filter:
907 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
908 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
909 1149
1150 dSP;
1151 ENTER; SAVETMPS;
1152 SAVESTACK_POS ();
1153 PUSHMARK (SP);
1154 EXTEND (SP, 2);
1155 PUSHs (tag_sv);
1156 PUSHs (sv);
1157
1158 PUTBACK;
1159 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1160 SPAGAIN;
1161
1162 if (SvTRUE (ERRSV))
1163 {
1164 SvREFCNT_dec (tag_sv);
1165 FREETMPS; LEAVE;
1166 ERR_ERRSV;
1167 }
1168
1169 if (count)
1170 {
1171 SvREFCNT_dec (tag_sv);
1172 SvREFCNT_dec (sv);
1173 sv = SvREFCNT_inc (POPs);
1174 }
1175 else
1176 {
910 AV *av = newAV (); 1177 AV *av = newAV ();
911 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
912 av_push (av, sv); 1179 av_push (av, sv);
913 1180
914 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
916 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1184 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1185 }
1186
1187 PUTBACK;
1188
1189 FREETMPS; LEAVE;
919 } 1190 }
920 break; 1191 break;
921 } 1192 }
922 1193
923 return sv; 1194 return sv;
930static SV * 1201static SV *
931decode_sv (dec_t *dec) 1202decode_sv (dec_t *dec)
932{ 1203{
933 WANT (1); 1204 WANT (1);
934 1205
935 switch (*dec->cur >> 5) 1206 switch (*dec->cur >> MAJOR_SHIFT)
936 { 1207 {
937 case 0: // unsigned int 1208 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938 return newSVuv (decode_uint (dec)); 1209 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939 case 1: // negative int 1210 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940 return newSViv (-1 - (IV)decode_uint (dec)); 1211 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941 case 2: // octet string 1212 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942 return decode_str (dec, 0); 1213 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943 case 3: // utf-8 string 1214 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944 return decode_str (dec, 1); 1215
945 case 4: // array 1216 case MAJOR_MISC >> MAJOR_SHIFT:
946 return decode_av (dec);
947 case 5: // map
948 return decode_hv (dec);
949 case 6: // tag
950 return decode_tagged (dec);
951 case 7: // misc
952 switch (*dec->cur++ & 31) 1217 switch (*dec->cur++ & MINOR_MASK)
953 { 1218 {
954 case 20: 1219 case SIMPLE_FALSE:
955#if CBOR_SLOW 1220#if CBOR_SLOW
956 types_false = get_bool ("Types::Serialiser::false"); 1221 types_false = get_bool ("Types::Serialiser::false");
957#endif 1222#endif
958 return newSVsv (types_false); 1223 return newSVsv (types_false);
959 case 21: 1224 case SIMPLE_TRUE:
960#if CBOR_SLOW 1225#if CBOR_SLOW
961 types_true = get_bool ("Types::Serialiser::true"); 1226 types_true = get_bool ("Types::Serialiser::true");
962#endif 1227#endif
963 return newSVsv (types_true); 1228 return newSVsv (types_true);
964 case 22: 1229 case SIMPLE_NULL:
965 return newSVsv (&PL_sv_undef); 1230 return newSVsv (&PL_sv_undef);
966 case 23: 1231 case SIMPLE_UNDEF:
967#if CBOR_SLOW 1232#if CBOR_SLOW
968 types_error = get_bool ("Types::Serialiser::error"); 1233 types_error = get_bool ("Types::Serialiser::error");
969#endif 1234#endif
970 return newSVsv (types_error); 1235 return newSVsv (types_error);
971 1236
972 case 25: 1237 case MISC_FLOAT16:
973 { 1238 {
974 WANT (2); 1239 WANT (2);
975 1240
976 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1241 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977 dec->cur += 2; 1242 dec->cur += 2;
978 1243
979 return newSVnv (ecb_binary16_to_float (fp)); 1244 return newSVnv (ecb_binary16_to_float (fp));
980 } 1245 }
981 1246
982 case 26: 1247 case MISC_FLOAT32:
983 { 1248 {
984 uint32_t fp; 1249 uint32_t fp;
985 WANT (4); 1250 WANT (4);
986 memcpy (&fp, dec->cur, 4); 1251 memcpy (&fp, dec->cur, 4);
987 dec->cur += 4; 1252 dec->cur += 4;
990 fp = ecb_bswap32 (fp); 1255 fp = ecb_bswap32 (fp);
991 1256
992 return newSVnv (ecb_binary32_to_float (fp)); 1257 return newSVnv (ecb_binary32_to_float (fp));
993 } 1258 }
994 1259
995 case 27: 1260 case MISC_FLOAT64:
996 { 1261 {
997 uint64_t fp; 1262 uint64_t fp;
998 WANT (8); 1263 WANT (8);
999 memcpy (&fp, dec->cur, 8); 1264 memcpy (&fp, dec->cur, 8);
1000 dec->cur += 8; 1265 dec->cur += 8;
1003 fp = ecb_bswap64 (fp); 1268 fp = ecb_bswap64 (fp);
1004 1269
1005 return newSVnv (ecb_binary64_to_double (fp)); 1270 return newSVnv (ecb_binary64_to_double (fp));
1006 } 1271 }
1007 1272
1008 // 0..19 unassigned 1273 // 0..19 unassigned simple
1009 // 24 reserved + unassigned (reserved values are not encodable) 1274 // 24 reserved + unassigned simple (reserved values are not encodable)
1275 // 28-30 unassigned misc
1276 // 31 break code
1010 default: 1277 default:
1011 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1278 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012 } 1279 }
1013 1280
1014 break; 1281 break;
1015 } 1282 }
1016 1283
1019} 1286}
1020 1287
1021static SV * 1288static SV *
1022decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1289decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1023{ 1290{
1024 dec_t dec = { }; 1291 dec_t dec = { 0 };
1025 SV *sv; 1292 SV *sv;
1026 STRLEN len; 1293 STRLEN len;
1027 char *data = SvPVbyte (string, len); 1294 char *data = SvPVbyte (string, len);
1028 1295
1029 if (len > cbor->max_size && cbor->max_size) 1296 if (len > cbor->max_size && cbor->max_size)
1043 if (dec.cur != dec.end && !dec.err) 1310 if (dec.cur != dec.end && !dec.err)
1044 dec.err = "garbage after CBOR object"; 1311 dec.err = "garbage after CBOR object";
1045 1312
1046 if (dec.err) 1313 if (dec.err)
1047 { 1314 {
1315 if (dec.shareable)
1316 {
1317 // need to break cyclic links, which would all be in shareable
1318 int i;
1319 SV **svp;
1320
1321 for (i = av_len (dec.shareable) + 1; i--; )
1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1323 sv_setsv (*svp, &PL_sv_undef);
1324 }
1325
1048 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1049 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1331 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1050 } 1332 }
1051 1333
1052 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1053 1335
1054 return sv; 1336 return sv;
1055} 1337}
1056 1338
1339/////////////////////////////////////////////////////////////////////////////
1340// incremental parser
1341
1342#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1343
1344// returns 0 for notyet, 1 for success or error
1345static int
1346incr_parse (CBOR *self, SV *cborstr)
1347{
1348 STRLEN cur;
1349 SvPV (cborstr, cur);
1350
1351 while (ecb_expect_true (self->incr_need <= cur))
1352 {
1353 // table of integer count bytes
1354 static I8 incr_len[MINOR_MASK + 1] = {
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 0, 0, 0, 0, 0, 0, 0, 0,
1357 0, 0, 0, 0, 0, 0, 0, 0,
1358 1, 2, 4, 8,-1,-1,-1,-2
1359 };
1360
1361 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1362 U8 m = *p & MINOR_MASK;
1363 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1364 I8 ilen = incr_len[m];
1365
1366 self->incr_need = self->incr_pos + 1;
1367
1368 if (ecb_expect_false (ilen < 0))
1369 {
1370 if (m != MINOR_INDEF)
1371 return 1; // error
1372
1373 if (*p == (MAJOR_MISC | MINOR_INDEF))
1374 {
1375 if (count >= 0)
1376 return 1; // error
1377
1378 count = 1;
1379 }
1380 else
1381 {
1382 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1383 count = -1;
1384 }
1385 }
1386 else
1387 {
1388 self->incr_need += ilen;
1389 if (ecb_expect_false (self->incr_need > cur))
1390 return 0;
1391
1392 int major = *p >> MAJOR_SHIFT;
1393
1394 switch (major)
1395 {
1396 case MAJOR_TAG >> MAJOR_SHIFT:
1397 ++count; // tags merely prefix another value
1398 break;
1399
1400 case MAJOR_BYTES >> MAJOR_SHIFT:
1401 case MAJOR_TEXT >> MAJOR_SHIFT:
1402 case MAJOR_ARRAY >> MAJOR_SHIFT:
1403 case MAJOR_MAP >> MAJOR_SHIFT:
1404 {
1405 UV len;
1406
1407 if (ecb_expect_false (ilen))
1408 {
1409 len = 0;
1410
1411 do {
1412 len = (len << 8) | *++p;
1413 } while (--ilen);
1414 }
1415 else
1416 len = m;
1417
1418 switch (major)
1419 {
1420 case MAJOR_BYTES >> MAJOR_SHIFT:
1421 case MAJOR_TEXT >> MAJOR_SHIFT:
1422 self->incr_need += len;
1423 if (ecb_expect_false (self->incr_need > cur))
1424 return 0;
1425
1426 break;
1427
1428 case MAJOR_MAP >> MAJOR_SHIFT:
1429 len <<= 1;
1430 case MAJOR_ARRAY >> MAJOR_SHIFT:
1431 if (len)
1432 {
1433 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1434 count = len + 1;
1435 }
1436 break;
1437 }
1438 }
1439 }
1440 }
1441
1442 self->incr_pos = self->incr_need;
1443
1444 if (count > 0)
1445 {
1446 while (!--count)
1447 {
1448 if (!AvFILLp (self->incr_count))
1449 return 1; // done
1450
1451 SvREFCNT_dec_NN (av_pop (self->incr_count));
1452 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1453 }
1454
1455 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1456 }
1457 }
1458
1459 return 0;
1460}
1461
1462
1057///////////////////////////////////////////////////////////////////////////// 1463/////////////////////////////////////////////////////////////////////////////
1058// XS interface functions 1464// XS interface functions
1059 1465
1060MODULE = CBOR::XS PACKAGE = CBOR::XS 1466MODULE = CBOR::XS PACKAGE = CBOR::XS
1061 1467
1069 1475
1070 types_true = get_bool ("Types::Serialiser::true" ); 1476 types_true = get_bool ("Types::Serialiser::true" );
1071 types_false = get_bool ("Types::Serialiser::false"); 1477 types_false = get_bool ("Types::Serialiser::false");
1072 types_error = get_bool ("Types::Serialiser::error"); 1478 types_error = get_bool ("Types::Serialiser::error");
1073 1479
1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1481
1074 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1075 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1076} 1486}
1077 1487
1078PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1079 1489
1080void CLONE (...) 1490void CLONE (...)
1099void shrink (CBOR *self, int enable = 1) 1509void shrink (CBOR *self, int enable = 1)
1100 ALIAS: 1510 ALIAS:
1101 shrink = F_SHRINK 1511 shrink = F_SHRINK
1102 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1103 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1104 allow_stringref = F_ALLOW_STRINGREF 1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1518 text_strings = F_TEXT_STRINGS
1519 validate_utf8 = F_VALIDATE_UTF8
1105 PPCODE: 1520 PPCODE:
1106{ 1521{
1107 if (enable) 1522 if (enable)
1108 self->flags |= ix; 1523 self->flags |= ix;
1109 else 1524 else
1115void get_shrink (CBOR *self) 1530void get_shrink (CBOR *self)
1116 ALIAS: 1531 ALIAS:
1117 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1118 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1119 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1120 get_allow_stringref = F_ALLOW_STRINGREF 1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1537 get_pack_strings = F_PACK_STRINGS
1538 get_text_keys = F_TEXT_KEYS
1539 get_text_strings = F_TEXT_STRINGS
1540 get_validate_utf8 = F_VALIDATE_UTF8
1121 PPCODE: 1541 PPCODE:
1122 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1123 1543
1124void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1544void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125 PPCODE: 1545 PPCODE:
1138 XPUSHs (ST (0)); 1558 XPUSHs (ST (0));
1139 1559
1140int get_max_size (CBOR *self) 1560int get_max_size (CBOR *self)
1141 CODE: 1561 CODE:
1142 RETVAL = self->max_size; 1562 RETVAL = self->max_size;
1563 OUTPUT:
1564 RETVAL
1565
1566void filter (CBOR *self, SV *filter = 0)
1567 PPCODE:
1568 SvREFCNT_dec (self->filter);
1569 self->filter = filter ? newSVsv (filter) : filter;
1570 XPUSHs (ST (0));
1571
1572SV *get_filter (CBOR *self)
1573 CODE:
1574 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1143 OUTPUT: 1575 OUTPUT:
1144 RETVAL 1576 RETVAL
1145 1577
1146void encode (CBOR *self, SV *scalar) 1578void encode (CBOR *self, SV *scalar)
1147 PPCODE: 1579 PPCODE:
1162 EXTEND (SP, 2); 1594 EXTEND (SP, 2);
1163 PUSHs (sv); 1595 PUSHs (sv);
1164 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1596 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165} 1597}
1166 1598
1599void incr_parse (CBOR *self, SV *cborstr)
1600 ALIAS:
1601 incr_parse_multiple = 1
1602 PPCODE:
1603{
1604 if (SvUTF8 (cborstr))
1605 sv_utf8_downgrade (cborstr, 0);
1606
1607 if (!self->incr_count)
1608 {
1609 self->incr_count = newAV ();
1610 self->incr_pos = 0;
1611 self->incr_need = 1;
1612
1613 av_push (self->incr_count, newSViv (1));
1614 }
1615
1616 do
1617 {
1618 if (!incr_parse (self, cborstr))
1619 {
1620 if (self->incr_need > self->max_size && self->max_size)
1621 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1622 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1623
1624 break;
1625 }
1626
1627 SV *sv;
1628 char *offset;
1629
1630 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1631 XPUSHs (sv);
1632
1633 sv_chop (cborstr, offset);
1634
1635 av_clear (self->incr_count);
1636 av_push (self->incr_count, newSViv (1));
1637
1638 self->incr_pos = 0;
1639 self->incr_need = self->incr_pos + 1;
1640 }
1641 while (ix);
1642}
1643
1644void incr_reset (CBOR *self)
1645 CODE:
1646{
1647 SvREFCNT_dec (self->incr_count);
1648 self->incr_count = 0;
1649}
1650
1651void DESTROY (CBOR *self)
1652 PPCODE:
1653 cbor_free (self);
1654
1167PROTOTYPES: ENABLE 1655PROTOTYPES: ENABLE
1168 1656
1169void encode_cbor (SV *scalar) 1657void encode_cbor (SV *scalar)
1658 ALIAS:
1659 encode_cbor = 0
1660 encode_cbor_sharing = F_ALLOW_SHARING
1170 PPCODE: 1661 PPCODE:
1171{ 1662{
1172 CBOR cbor; 1663 CBOR cbor;
1173 cbor_init (&cbor); 1664 cbor_init (&cbor);
1665 cbor.flags |= ix;
1174 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1666 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175 XPUSHs (scalar); 1667 XPUSHs (scalar);
1176} 1668}
1177 1669
1178void decode_cbor (SV *cborstr) 1670void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines